Participant Testimonials
The workshop provided a high level overview of what the experts were currently thinking and working on. The ideas on profile likelihoods were very valuable. I will be going back to the talks on the website many times in the next few months. I am hopeful that some work (and a paper) on negative weights will result.
The workshop was excellent and strongly impacted and will impact my current research. My current research is the search for dark matter via indirect detection using the Fermi gamma ray space telescope. The attendees of the meeting were not the usual scientists that I would interact with, and their focus on statistics of discovery made for a very special and unusual conference. This conference peaked my interest in Bayesian statistics, and probably will lead to a graduate course that I will teach on the subject this Fall. I am sure I will understand the subject much better after teaching it to our KIPAC graduate students.
KIPAC, SLAC, Stanford University
I had a very stimulating week at the meeting. BIRS is an excellent venue for workshops of this scale which are, in my opinion, the most fruitful scientific meetings I attend.
Many thanks for hosting an extremely productive workshop in an exceptionally beautiful and stimulating environment. The generosity and hospitality of BIRS has been outstanding. My colleagues Eilam Gross, Ofer Vitells, Kyle Cranmer and I had just completed a paper on the application of asymptotic distributions likelihood ratio statistics for discoveries at the LHC, and the workshop was the perfect venue to present the work and get important feedback. The other contributions as well were very interesting and of direct benefit to the efforts at the LHC to prepare for new discoveries.
This was my second time at BIRS, and both times have been very productive and enjoyable. On this trip, I started two new collaborative projects that I expect to bear fruit in the coming years. I also sat down with a participant who showed me how to prove a conjecture that I had made at the conference! So not only did it inspire creativity, BIRS catalyzed real scientific results.
This was a superb workshop for me. I work directly in this business and we have many high-stakes results being produced at the Tevatron as it collects more data than ever before, opening up our sensitivity to processes we have not been able to test before. It's always a pleasure to talk with people at the LHC who are preparing to confront their data and models with statistical techniques. We have in the past and will continue to learn from each other. It's also quite a pleasure to talk with statisticians on the subject. Some things we are getting consensus on (I think we are settling on a consistent handling of the look-elsewhere effect, for example), and there are other open problems which may not be statistical in nature but which require us to do more work. The organization was outstanding, and the staff very cheerful and helpful.
... thanks to Wynne, to Brenda and Brent and anyone else responsible for what our participants all tell me was a great workshop on statistical issues in discovery claims in particle and astrophysics. The atmosphere produced wonderful vigorous interactions between the three groups and we are all grateful for the opportunity and for the way we were looked after all the way along.
This was an excellent week, both providing fresh insights into an area I am already familiar with and completely different perspectives from the neighbouring astrophysics challenges. The relaxed BIRS atmosphere made for a great environment for informal discussion, which, as usual, was the most important part of the workshop. The single most urgent problem we face, the statistical methods to be used to treat the LHC data, took a big step forward both in a predictable direction (the profile likelihood discussions) and in one I did not foresee (the look elsewhere effect) This second topic greatly benefited from the varied perspectives of the mixed group assembled for the meeting.
Particle Physics, Rutherford Appleton Lab
The meeting on statistical issues pertaining to discovery was one of the most productive meetings I have attended during the past few years. A broad range of ideas were discussed, many of which directly inform my own research in high energy physics. I report briefly on two topics of particular interest to me. I had the opportunity to discuss, with Kyle Cranmer, the implementation of reference priors within the CERN statistical software package RooStats and to sketch out a longer-term project. Reference priors are the cornerstone of a state-of-the-art Bayesian methodology called reference analysis, in which some work by Luc Demortier, Supriya Jain and me was recently completed. Following my discussions with Kyle, I have a much clearer picture of how I should proceed over the next few months. The second topic in which progress was made is that of parton distribution functions (PDFs). Parton distribution functions describe, at a given resolution, the structure of particles such as protons. These functions are a critical input in theoretical predictions of the results of particle collisions at accelerators, such as those occurring now at the Large Hadron Collider (LHC) at CERN. In several discussion with theorists Jon Pumplin and Robert Thorne, a consensus emerged about the utility (indeed, necessity) of performing end-to-end simulations of the statistical methods they use to infer the PDFs from fits to particle collision data. My colleagues recognize that end-to-end simulations are needed if we are to make progress in clarifying currently obscure aspects of their fitting procedures. I regard the launching of a fresh attack on the problem of understanding PDF uncertainties as an important success of the BIRS workshop. There is promise of substantial progress over the next few months. My hope is that Jon and Robert will be able to report as much at the PhyStat 2011 workshop at CERN. This was an enjoyable and productive workshop, made all the more so by the gracious and helpful BIRS staff.
Thanks very much for hosting the Statistical Issues workshop! It provided for a great deal of forward progress on my research area of Parton Distribution Functions. It was particularly useful to have two of the people present (Robert Thorne and myself) who are most involved in the current parton distribution efforts, and who represent different collaborations doing this work, along with a number of high energy physicists,and along with statisticians who were not familiar with this area before the workshop, but who were none-the-less able to quickly grasp the essentials and come up with some new approaches which I am hopeful will turn out to be very productive.
The workshop was very useful for my work as it exposed me to cross-disciplinary issues faced by colleagues in particle physics. I also very much appreciated the opportunity to interact with statisticians. The opportunities for networking and starting up collaborative work were excellent.
Physics Dept, Astrophysics Group, Imperial College London
